R ggplot2：在多个变量的相同图中添加均值和标准差

Question

I have 4 vectors of 16 values each; each values is the mean of an item, and I have the same 16 items across 4 datasets.

I use ggplot2 to plot these means: here a reproducible example.

library("ggplot2")
library("dplyr")    

means <- as.data.frame(cbind(rnorm(16),rnorm(16), rnorm(16), rnorm(16)))
means <- mutate(means, id = rownames(means))
colnames(means)<-c("1", "2", "3", "4", "Symptoms")
means_long <- melt(means, id="Symptoms")
means_long$Symptoms <- as.numeric(means_long$Symptoms)
names(means_long)[2] <- "Datasets"

ggplot(data=means_long, aes(x=Symptoms, y=value, colour=Datasets)) +
      geom_line() +
      geom_point(shape = 21, fill = "white", size = 1.5, stroke = 1) +
      xlab("Symptoms") + ylab("Means") +
      scale_y_continuous() + 
      scale_x_continuous(breaks=c(1:16)) +
      theme_bw() +
      theme(panel.grid.minor=element_blank()) +
      coord_flip()

Now, I have 4 further vectors, which are the standard deviations of the 16 items for the 4 datasets. I want to plot them into the same plot. The data are in the same format as above, so it's virtually the same code.

I want the standard deviations in the same plot as the means, using the same colors but different line types (so dataset 1 mean is red, dataset 1 standard deviation is dashed), and in the best case a legend that differentiates both by dataset (as I have currently) in addition to "mean" vs "standard deviation" for the lines and dashed lines.

Thank you for your help!

Answer 1

Does this help?

To make it not look super ugly, I made all the random mean values positive, and then just made the example standard deviations negative. The way of plotting the values on the same graph is to feed in the datasets separately to each geom, rather than defining in initial ggplot() function.

Let me know if this isn't what you were thinking:

library("ggplot2")
library("dplyr")    

means <- as.data.frame(abs(cbind(rnorm(16),rnorm(16), rnorm(16), rnorm(16))))
means <- mutate(means, id = rownames(means))
colnames(means)<-c("1", "2", "3", "4", "Symptoms")
means_long <- reshape2::melt(means, id="Symptoms")
means_long$Symptoms <- as.numeric(means_long$Symptoms)
names(means_long)[2] <- "Datasets"


sds_long <- means_long
sds_long$value <- -sds_long$value

ggplot() +
  geom_line(aes(x=Symptoms, y=value, colour=Datasets), lty=1, data=means_long) +
  geom_point(aes(x=Symptoms, y=value, colour=Datasets), data=means_long, shape = 21, fill = "white", size = 1.5, stroke = 1) +
  geom_line(aes(x=Symptoms, y=value, colour=Datasets), lty=2, data=sds_long) +
  geom_point(  aes(x=Symptoms, y=value, colour=Datasets), data=sds_long, shape = 21, fill = "white", size = 1.5, stroke = 1) +
  xlab("Symptoms") + ylab("Means") +
  scale_y_continuous() + 
  scale_x_continuous(breaks=c(1:16)) +
  theme_bw() +
  theme(panel.grid.minor=element_blank()) +
  coord_flip()

#

To answer your legend query. In short, I think this is very hard because the same mapping aesthetic is being used with both datasets.

However, using the code from this answer - I did the following. The idea is to get the legend from two plots only plotting means/sds and then adding those legends to a version of the plot with no legend. It could be adapted so you position the legends more manually...

 ### Step 1 # Draw a plot with the colour legend p1 <- ggplot() + geom_line(aes(x=Symptoms, y=value, colour=Datasets), lty=1, data=means_long) + geom_point(aes(x=Symptoms, y=value, colour=Datasets), data=means_long, shape = 21, fill = "white", size = 1.5, stroke = 1) + scale_color_manual(name = "Means",values=c("red","blue", "green","pink")) + coord_flip()+ theme_bw() + theme(panel.grid.minor=element_blank()) + theme(legend.position = "top") # Extract the colour legend - leg1 library(gtable) leg1 <- gtable_filter(ggplot_gtable(ggplot_build(p1)), "guide-box") ### Step 2 # Draw a plot with the size legend p2 <- ggplot() + geom_line(aes(x=Symptoms, y=value, color=Datasets), lty=2, data=sds_long) + geom_point( aes(x=Symptoms, y=value, color=Datasets), data=sds_long, shape = 21, fill = "white", size = 1.5, stroke = 1) + coord_flip()+ theme_bw() + theme(panel.grid.minor=element_blank()) + scale_color_manual(name = "SDs",values=c("red","blue", "green","pink")) # Extract the size legend - leg2 leg2 <- gtable_filter(ggplot_gtable(ggplot_build(p2)), "guide-box") # Step 3 # Draw a plot with no legends - plot p3<-ggplot() + geom_line(aes(x=Symptoms, y=value, colour=Datasets), lty=1, data=means_long) + geom_point(aes(x=Symptoms, y=value, colour=Datasets), data=means_long, shape = 21, fill = "white", size = 1.5, stroke = 1) + geom_line(aes(x=Symptoms, y=value, color=Datasets), lty=2, data=sds_long) + geom_point( aes(x=Symptoms, y=value, color=Datasets), data=sds_long, shape = 21, fill = "white", size = 1.5, stroke = 1) + xlab("Symptoms") + ylab("Means") + scale_y_continuous() + scale_x_continuous(breaks=c(1:16)) + theme_bw() + theme(panel.grid.minor=element_blank()) + coord_flip()+ scale_color_manual(values=c("red","blue", "green","pink")) + theme(legend.position = "none") ### Step 4 # Arrange the three components (plot, leg1, leg2) # The two legends are positioned outside the plot: # one at the top and the other to the side. library(grid) plotNew <- arrangeGrob(leg1, p3, heights = unit.c(leg1$height, unit(1, "npc") - leg1$height), ncol = 1) plotNew <- arrangeGrob(plotNew, leg2, widths = unit.c(unit(1, "npc") - leg2$width, leg2$width), nrow = 1) grid.newpage() grid.draw(plotNew) 

Answer 2

I suggest an implementation that you only need 1 single dataframe to plot. Plus you don't need to tweak your code much, but you are still be able to distinguish datasets (ie, 1, 2, 3, 4) and types of values (eg, mean, sd).

library("ggplot2")
library("dplyr")    

# Means
means <- as.data.frame(cbind(rnorm(16),rnorm(16), rnorm(16), rnorm(16)))
means <- mutate(means, id = rownames(means))
colnames(means)<-c("1", "2", "3", "4", "Symptoms")
means_long <- melt(means, id="Symptoms")
means_long$Symptoms <- as.numeric(means_long$Symptoms)
names(means_long)[2] <- "Datasets"

# Sd
sds_long <- means_long
sds_long$value <- -sds_long$value

################################################################################
# Add "Type" column to distinguish means and sds
################################################################################
type <- c("Mean")
means_long <- cbind(means_long, type)

type <- c("Sd")
sds_long <- cbind(sds_long, type)

merged <- rbind(means_long, sds_long)

colnames(merged)[4] <- "Type"

################################################################################
# Plot
################################################################################
ggplot(data = merged) +
  geom_line(aes(x = Symptoms, y = value, col = Datasets, linetype = Type)) +
  geom_point(aes(x = Symptoms, y = value, col = Datasets), 
             shape = 21, fill = "white", size = 1.5, stroke = 1) +
  xlab("Symptoms") + ylab("Means") +
  scale_y_continuous() + 
  scale_x_continuous(breaks=c(1:16)) +
  theme_bw() +
  theme(panel.grid.minor=element_blank()) +
  coord_flip()